Hydraulic support support imbalance grading early warning method for super-high and super-long working face

Abstract

The application provides a kind of hydraulic support supporting imbalance grading early warning method for super-high and super-long working face, and relates to the technical field of coal mine safety.The method comprises the following steps: obtaining multi-source sensing data of the hydraulic support and constructing a coupled feature vector;establishing a state space model with internal support mechanical state variables as state vectors and measurable physical quantities as observation variables;using an extended Kalman filter to recursively estimate the state vector to obtain the state estimate and its covariance;calculating the imbalance index based on the state estimate, and determining the support state grade according to the grading determination rule.The application maps the multi-source measurable information to the internal mechanical state of the support which cannot be directly measured, improves the imbalance identification basis from the threshold judgment of apparent measurable variables to the comprehensive evaluation of internal mechanical state, and determines the imbalance grade by reconstructing the state, which can identify the support imbalance precursor in advance when a single apparent variable is not yet out of range, improve the early warning lead time, and reduce the risk of false positives and false negatives.

Description

Technical Field

[0001] This application relates to the field of coal mine safety technology, and in particular to a method for graded early warning of imbalance in hydraulic support for ultra-high and ultra-long working faces. Background Technology

[0002] Hydraulic supports are the core support equipment in fully mechanized mining faces, and their support status directly affects roof control and mining safety. With the promotion of ultra-long working faces and high-intensity mining, supports are prone to imbalance phenomena such as eccentric loading, abnormal posture, inter-support compression, and degradation of support stiffness under working conditions such as periodic pressure, floor undulation, and asynchronous movement between supports.

[0003] In related technologies, monitoring methods often employ single-variable thresholds or simple multi-variable logical thresholds. This involves setting upper and lower limits for parameters such as column pressure, top beam inclination, and displacement, triggering an alarm when these limits are exceeded. These methods only assess seemingly measurable variables and struggle to identify early signs of imbalance, such as continuously increasing off-center loads before exceeding limits, simultaneous increases in load and attitude change rates before obvious attitude abnormalities, and rising inter-frame constraint reaction forces before displacement reaches the fault threshold. Furthermore, while some technologies utilize multiple sensors for support attitude fusion or automatic pressure regulation control, they cannot monitor directly measurable internal mechanical states. When control is triggered by a single threshold, misadjustment, delayed adjustment, or overadjustment can easily occur under complex operating conditions, resulting in coarse warning grading and unclear safety control boundaries. Therefore, a method is urgently needed that can fuse attitude and load data from multiple sources to reconstruct the directly measurable internal support state and provide graded warnings based on this reconstructed state. Summary of the Invention

[0004] This application aims to at least partially address one of the technical problems in the related art.

[0005] Therefore, the first aspect of this application proposes a graded early warning method for imbalance of hydraulic support in ultra-high and ultra-long working faces, comprising the following steps:

[0006] Acquire multi-source information for hydraulic supports used in ultra-high and ultra-long working faces. The multi-source information includes: left column pressure, right column pressure, top beam pitch angle, top beam tilt angle, base pitch angle, base tilt angle, balance jack pressure, pushing displacement, adjacent support pushing displacement, and relative displacement between supports. Based on the multi-source information, a coupled feature vector is constructed, which includes: column load asymmetric component, top beam relative to base attitude deviation component, attitude-load coordinated change rate component, inter-frame constraint component, and load disturbance component. Using the internal support resultant force, eccentric load moment, attitude-load coupling degree, inter-frame constraint reaction force and support stiffness degradation coefficient as state vectors, and the column pressure, eccentric load equivalent moment, relative attitude deviation, shoving-constraint coupling displacement and balancing jack pressure as observation vectors, a support state space model including state equations and observation equations is established. An extended Kalman filter is used to recursively solve the support state space model based on the multi-source information and the coupling feature vector to obtain the estimated value of the state vector and the covariance estimate of the state estimation error. The imbalance index of the hydraulic support is determined based on the estimated value of the state vector; Based on the imbalance index and the preset support status grading rules, the support status level of the hydraulic support is determined and an early warning is issued.

[0007] The second aspect of this application provides a graded early warning device for imbalance of hydraulic support for ultra-high and ultra-long working faces, comprising: The acquisition module is used to acquire multi-source information of the hydraulic support for ultra-high and ultra-long working face. The multi-source information includes: left column pressure, right column pressure, top beam pitch angle, top beam tilt angle, base pitch angle, base tilt angle, balance jack pressure, pushing displacement, adjacent support pushing displacement and relative displacement between supports. The feature extraction module is used to construct a coupled feature vector based on the multi-source information. The coupled feature vector includes: column load asymmetric component, top beam relative to base attitude deviation component, attitude-load coordinated change rate component, inter-frame constraint component, and load disturbance component. The module is used to establish a support state space model that includes state equations and observation equations, with the internal support resultant force, eccentric load moment, attitude-load coupling degree, inter-frame constraint reaction force and support stiffness degradation coefficient as state vectors, and column pressure, eccentric load equivalent moment, relative attitude deviation, push-constraint coupling displacement and balancing jack pressure as observation vectors. The solution module is used to recursively solve the support state space model using an extended Kalman filter based on the multi-source information and the coupling feature vector, to obtain the estimated value of the state vector and the covariance estimate of the state estimation error. A determination module is used to determine the imbalance index of the hydraulic support based on the estimated value of the state vector; The graded early warning module is used to determine the support status level of the hydraulic support and issue an early warning based on the imbalance index and the preset support status grading judgment rules.

[0008] A third aspect of this application provides an electronic device, including: a processor, and a memory communicatively connected to the processor; The memory stores computer-executed instructions; The processor executes computer execution instructions stored in the memory to implement the method described in the first aspect above.

[0009] A fourth aspect of this application provides a computer-readable storage medium, characterized in that the computer-readable storage medium stores computer-executable instructions, which, when executed by a processor, are used to implement the method described in the first aspect above.

[0010] The method for graded early warning of support imbalance in hydraulic supports for ultra-high and ultra-long working faces provided in this application constructs a coupled feature vector from multi-source information characterizing the support state, and establishes a state-space model with the internal mechanical state quantities of the support that cannot be directly measured as the state vector and the measurable physical quantities as the observables. An extended Kalman filter is used to recursively estimate the state vector, realizing online reconstruction of the internal support state of the support. Based on the reconstructed state, an imbalance index is calculated and an early warning is issued according to the graded judgment rules. This elevates the basis for imbalance identification from threshold judgment of apparent measurable variables to a comprehensive assessment of the internal mechanical state. It can detect early signs of support imbalance before a single sensor reading reaches the traditional alarm threshold, increasing the lead time for anomaly identification and reducing the risk of false alarms and missed alarms caused by sensor noise or instantaneous disturbances.

[0011] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0012] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein: Figure 1 A flowchart illustrating a graded early warning method for imbalance in hydraulic support for ultra-high and ultra-long working faces, provided for an embodiment of this application; Figure 2 A schematic diagram of a hydraulic support provided in an embodiment of this application; Figure 3 A system architecture diagram of a multi-source sensor array and a downhole edge computing unit provided in this application embodiment; Figure 4 This application provides a schematic diagram of multi-source information hardware synchronization, timestamp verification, and unified sampling in an embodiment. Figure 5 A schematic diagram illustrating the construction of a coupled feature vector provided in an embodiment of this application; Figure 6 A schematic diagram illustrating the mapping relationship between a support state-space model and observation equations provided in this application embodiment; Figure 7 A flowchart of the state reconstruction process of an extended Kalman filter (EKF) with online parameter identification is provided for embodiments of this application. Figure 8 A schematic diagram of a graded early warning decision tree for support imbalance provided in this application embodiment; Figure 9 This is a diagram illustrating a hierarchical early warning and security control interface output architecture proposed in an embodiment of this application. Figure 10 This is a flowchart of a non-stationary operating condition sudden change detection method proposed in an embodiment of this application; Figure 11 A flowchart illustrating another method for graded early warning of imbalance in hydraulic support for ultra-high and ultra-long working faces provided in this application embodiment; Figure 12 This is a comparison diagram of early warning effects under a typical working condition proposed in an embodiment of this application; Figure 13 This is a schematic diagram of a hydraulic support imbalance classification and early warning device for ultra-high and ultra-long working faces, provided as an embodiment of this application. Detailed Implementation

[0013] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, and should not be construed as limiting this application.

[0014] Specifically, the following describes a method for graded early warning of imbalance in hydraulic support for ultra-high and ultra-long working faces, based on embodiments of this application, with reference to the accompanying drawings.

[0015] Figure 1 This is a flowchart illustrating a graded early warning method for imbalance in hydraulic support for ultra-high and ultra-long working faces, provided as an embodiment of this application. Figure 1 As shown, the method for graded early warning of imbalance in hydraulic support for ultra-high and ultra-long working faces may include the following steps: Step 101: Obtain multi-source information of the hydraulic support for the ultra-high and ultra-long working face. The multi-source information includes: left column pressure, right column pressure, top beam pitch angle, top beam tilt angle, base pitch angle, base tilt angle, balance jack pressure, pushing displacement, adjacent support pushing displacement, and relative displacement between supports.

[0016] In some embodiments of this application, multi-source information can be collected by a multi-source sensor group deployed on a hydraulic support. Figure 2 This is a schematic diagram of a hydraulic support provided in an embodiment of this application. Figure 3This document presents a system architecture diagram of a multi-source sensor array and a downhole edge computing unit, as provided in an embodiment of this application. Figure 2 and Figure 3 As shown, the hydraulic support 400 may include a top beam 410, a column 420, a base 430, a balancing jack 440, and a pushing mechanism 450. A column pressure sensor 510 is installed in the hydraulic circuit of the column 420 to obtain the pressure p of the left column. f and right column pressure p r A dual-axis tilt sensor 520 is installed on the top beam 410 to obtain the top beam pitch angle θ. p and the tilt angle θ of the top beam r A dual-axis base tilt sensor 550 is installed on the base 430 to obtain the base pitch angle θ. bp and base tilt angle θ br A jack pressure sensor 530 is installed on the jack 440 to obtain the jack pressure p. b A displacement sensor 540 is installed on the pushing mechanism 450 to obtain the pushing displacement Δs(n) and the pushing displacement Δs(n+1) of the adjacent support. An inter-support relative displacement sensor 560 is installed between adjacent hydraulic supports to obtain the inter-support relative displacement d. n The downhole edge computing unit can use PTP or IEEE 1588 time synchronization signals to provide a unified trigger reference for each sensor.

[0017] In one example, the hydraulic support imbalance classification early warning method for ultra-high and ultra-long working faces proposed in this application embodiment can be deployed in the downhole edge computing unit and connected to the working face centralized control system, support controller or human-machine interface through the data interface. It does not require the installation of a special off-center load torque sensor or inter-support force sensor, and is easy to integrate with the existing electro-hydraulic control system.

[0018] As an example, the parameter settings for a multi-source sensor array can be found in Table 1.

[0019] Table 1

[0020] In some embodiments of this application, each sensor can be connected to the downhole edge computing unit 600 via a mining intrinsically safe communication bus or an industrial Ethernet. The downhole edge computing unit 600 has hardware time synchronization capabilities, sending a unified trigger signal to each sensor. Each sensor writes a hardware timestamp T during sampling. i The timing synchronization module has 620 pairs of T... i With unified trigger time T trig The deviation Δt between i Perform a check when |T i Ttrig | Greater than the preset deviation upper limit Δt max At that time, the samples are interpolated and resampled or removed, and finally the data under a unified time reference is output. All signals are finally aligned to a unified output reference of 10Hz to 200Hz.

[0021] Figure 4 This diagram illustrates a multi-source information hardware synchronization, timestamp verification, and unified sampling method provided in this application embodiment. For example, pressure, tilt, and displacement signals can be uniformly aligned to 100Hz. For sensors with weak communication capabilities, data can be collected first at the original sampling frequency, and then aligned by interpolation from the edge computing unit. This time synchronization design ensures that the attitude change rate and load change rate have the same time reference, avoiding subsequent z3 calculation distortion due to phase differences.

[0022] Step 102: Construct a coupled feature vector based on multi-source information. The coupled feature vector includes: column load asymmetric component, top beam attitude deviation component relative to base, attitude-load coordinated change rate component, inter-frame constraint component, and load disturbance component.

[0023] In some embodiments of this application, the coupling feature vector can be constructed using the following formula: z1=|p f p r | / p rated z2=√[(θ p θ bp ) 2 +(θ r θ br ) 2 ] / θ ref z3=α1·|dz1 / dt|+α2·|dz2 / dt|+α3·|d(p b / p rated ) / dt| z4=[|Δs(n) Δs(n+1)|+κ d ·|d n |] / Δs design z5=σ τ (p f +p r ) / (2p rated ) Where z1 is the asymmetric component of the column load, z2 is the attitude deviation component of the top beam relative to the base, z3 is the attitude-load coordinated rate of change component, z4 is the inter-frame constraint component, z5 is the load disturbance component, and pf For the pressure on the left column, p r For the pressure on the right column, θ p Let θ be the pitch angle of the top beam. r Let θ be the tilt angle of the top beam. bp Let θ be the pitch angle of the base. br p is the base tilt angle. b To balance the pressure of the jacks, Δs(n) is the displacement, Δs(n+1) is the displacement of the adjacent support, and d n p represents the relative displacement between the frames. rated For the rated column pressure, θ ref As the baseline for attitude tolerance, Δs design The design step size is τ, where τ is the length of the sliding window, and σ is the sliding window size. τ (·) represents the standard deviation within the window τ, where α1, α2, α3, κ are constants. d These are calibrable weighting coefficients.

[0024] Among them, the asymmetric component z1 of the column load is used to represent the degree of pressure imbalance between the left and right columns, which is different from directly using p. f or p r Unlike other values, z1 can reflect the relative off-center loading trend between the left and right columns, even if p f and p r None of them exceeded the univariate stress threshold, and z1 may continue to rise; The top beam relative to the base attitude deviation component z2 is used to represent the comprehensive attitude deviation of the top beam relative to the base. By introducing the base attitude as a reference, the interference of the overall tilt of the base plate or the overall tilt of the support on the attitude judgment of the top beam can be reduced. The attitude-load coordinated change rate component z3 is used to represent the dynamic coupling trend between load change, attitude change and balancing jack pressure change. When z1 and z2 have not reached the severe threshold but change rapidly in sync, z3 can characterize the imbalance development trend in advance. The inter-frame constraint component z4 is used to represent the degree of coordination between the current frame and the adjacent frames and the degree of inter-frame compression. This component can reflect constraint anomalies caused by asynchronous movement, lateral compression between frames, and base misalignment. The load disturbance component z5 is used to represent the degree of fluctuation of the support load within the sliding window τ. This component can help identify non-steady working conditions such as periodic pressure, roof rupture, or fault disturbance.

[0025] Figure 5 This is a schematic diagram illustrating the construction of a coupled feature vector according to an embodiment of this application. In some embodiments of this application, such as... Figure 5As shown, outlier removal, low-pass filtering, and dimensional normalization can be performed on each synchronous data stream. Outlier removal can be achieved by combining physical range constraints and Hampel filtering. Low-pass filtering can employ first-order inertial filtering or finite impulse response filtering. Normalization parameters can be selected from the support's rated pressure, design attitude deviation, and design push-off step distance.

[0026] Step 103: Using the internal support resultant force, eccentric load moment, attitude-load coupling degree, inter-frame constraint reaction force and support stiffness degradation coefficient as state vectors, and the column pressure, eccentric load equivalent moment, relative attitude deviation, push-constraint coupling displacement and balancing jack pressure as observation vectors, establish a support state space model including state equations and observation equations.

[0027] Define the state vector as x = [x1, x2, x3, x4, x5] T .

[0028] Where x1 is the internal support resultant force, which can be obtained by observing the combination of the pressure of the left and right columns and the effective area of ​​the columns. It represents the equivalent support resultant force borne by the top beam through the columns and is used to judge the change in support strength. x2 is the eccentric load moment, which can be estimated by the pressure difference between the left and right columns, the equivalent force arm, and the relative attitude deviation. It represents the equivalent moment caused by the pressure difference between the left and right columns and the attitude deviation, and is used to identify eccentric load imbalance. x3 represents the attitude-load coupling degree, which can be estimated by the sliding correlation and coordinated change rate of z1 and z2. It represents the dynamic correlation between attitude change and load change and is used to identify linkage anomalies. x4 is the inter-frame constraint reaction force, which can be estimated from the difference in displacement between this frame and the adjacent frame, the relative displacement between frames, and the equivalent constraint stiffness. It represents the lateral or displacement constraint exerted by the adjacent support on this frame and is used to identify inter-frame squeezing / displacement anomalies. x5 is the support stiffness degradation coefficient, which can be estimated from information such as load disturbance, duration of off-center load, and the relationship between the resultant force and displacement of the support. It represents the degree of degradation of the current support stiffness relative to the design stiffness and is used to identify the risk of chronic instability.

[0029] In some embodiments of this application, the state equation may include the following recursive relation: x1(k)=x1(k-1)+a 11 ·Δp b (k-1)+a 12 ·z5(k-1)+w1(k-1) x2(k)=a 21 ·x2(k-1)+a 22 ·A p ·L c ·(p f p r (k-1)+a 23 ·z2(k-1)+w2(k-1) x3(k)=a 31 ·x3(k-1)+a 32 ·corr τ (dz1 / dt, dz2 / dt) + w3(k-1) x4(k)=a 41 ·x4(k-1)+a 42 ·k-1 c ·[Δs(n) Δs(n+1)+κ d ·d n ](k-1)+w4(k-1) x5(k) = x5(k-1) a 51 ·max{0,z5(k-1) z 5,0} a 52 ·max{0,|x2(k-1)| M0}+w5(k-1) Where k is the current time, k-1 is the previous time, x1 is the resultant force of internal support, x2 is the eccentric load moment, x3 is the attitude-load coupling degree, x4 is the inter-frame constraint reaction force, x5 is the support stiffness degradation coefficient, and p f For the pressure on the left column, p r The pressure on the right column is Δs(n), the displacement is Δs(n+1), and the displacement of the adjacent support is d. n For the relative displacement between the frames, Δ(p) b To balance the pressure changes of the jack, corr τ Let k be the sliding correlation coefficient for length τ. c For the equivalent constraint stiffness between frames, κ d For calibrable weighting coefficients, z 5,0 As the load disturbance reference, M0 is the design off-center load moment threshold, a 11 a 12 a 21 a 22 a 23 a 31 a 32 a 41 a 42 a 51 a 52For online identification or offline calibration of parameters, w1, w2, w3, w4, and w5 represent the process noise of each state component, used to characterize unmodeled dynamics and system disturbances. Their statistical properties are described by the process noise covariance Q(k), A p L is the effective pressure-bearing area of ​​the column. c This refers to the center distance of the support column or the equivalent force arm. This application elevates the detection of support imbalance from superficial monitoring to internal state reconstruction. By reconstructing the state vector, the identification of the support state more closely approximates the true mechanical state of the support.

[0030] Define the observation vector as y = [y1, y2, y3, y4, y5]. T The observation vector can transform pressure, attitude, surfacing, and inter-frame displacement into equivalent observations related to the internal state. Here, y1 is the column pressure, y2 is the eccentric load equivalent torque, y3 is the relative attitude deviation, y4 is the surfacing-constraint coupled displacement, and y5 is the balancing jack pressure. In some embodiments of this application, the actual value of the observation vector can be calculated using the following formula: y1=A p ·(p f +p r ) y2=A p ·L c ·(p f p r ) y3=z2 y4=Δs(n) Δs(n+1)+κ d ·d n y5=p b / p rated The meaning of the parameters in the formula for the actual value of the observation vector has been given above and will not be repeated here.

[0031] Step 104: Using an extended Kalman filter, the support state-space model is recursively solved based on multi-source information and coupled eigenvectors to obtain the estimated value of the state vector and the covariance estimate of the state estimation error.

[0032] Figure 6 This diagram illustrates the mapping relationship between a support state-space model and an observation equation, as provided in an embodiment of this application. In some embodiments of this application, the solution process for the extended Kalman filter at each sampling time may include the following steps S11-S16: S11, based on the state equation and the state vector of the previous time step, calculate the initial estimate of the state vector at the current time step.

[0033] The state vector x(k-1) from the previous time step, the multi-source information, and the coupled feature vector are substituted into the state equation in step 103, and the calculated result is used as the initial estimate of the state vector at the current time step. (k|k 1). The initial estimate of the state vector can be expressed as: (k|k 1) = f( (k 1|k 1), u(k 1), z(k 1), β(k 1))+w(k),u(k) 1) This refers to the control input from the previous moment, including the control input u of the balancing jack. b (e.g., the pressure adjustment amount Δp of the balancing jack) b ), z(k 1) represents the coupling feature vector from the previous time step, β(k) 1) represents the online identification parameter vector from the previous time step, β(k) = [a 11 ,a 12 ,...,c 11 ,c 12 ,...,k c ,κ d ], w(k) is the process noise.

[0034] S12, based on the covariance of the state estimation error at the previous time step and the process noise covariance, calculate the initial estimate of the covariance of the state estimation error at the current time step.

[0035] The covariance of the state estimation error describes the uncertainty of the estimated state vector. A larger value in the covariance P(k) indicates a lower confidence level in the current state vector estimate, and a less reliable estimate; a smaller value indicates a more accurate estimate.

[0036] For example, the initial estimate of the covariance of the current state estimation error can be calculated using the following formula: P(k|k 1) = F(k)P(k 1|k 1)F(k) T + Q(k) Wherein, P(k) 1|k 1) is the covariance of the state estimation error at the previous time step, F(k) is the Jacobian matrix of the state equation with respect to the state vector, and Q(k) is the process noise covariance.

[0037] S13, based on multi-source information, coupled feature vectors, and observation equations, determines the theoretical and actual values ​​of the observation vectors.

[0038] The observation equation may include the following observation relationships: y1=c 11 ·x1+v1 y2=c 21 ·x2+c 22 ·z2+v2 y3=c 31 ·x3+c 32 ·(1 x5)+v3 y4=c 41 ·x4+c 42 ·dx4 / dt+v4 y5=c 51 ·x2+c 52 ·z2+c 53 ·u b +v5 Among them, u b To balance the jack control input, c 11 -c 53 For online identification or offline calibration parameters, they can be determined by the stent's factory parameters, calibration tests, and historical operating data.

[0039] Substituting the initial estimate of the current state vector calculated in S11 into x in the above observation relationship, we can obtain the theoretical value of the observation vector. This represents the theoretically measurable observation vector when the internal state that cannot be directly measured is x (the initial estimate of the state vector) deduced from multi-source information. Specifically, the initial estimate of the current state vector... (k|k 1) The observation equation for calculating the theoretical value of the observed vector can be expressed as h( (k|k 1), z(k),β(k))+ v(k), where v(k) is the observation noise.

[0040] S14. Determine the innovation residual based on the theoretical and actual values ​​of the observed vector.

[0041] The innovation residual e(k) is the theoretical value h(k) of the observed vector. (k|k The difference between 1), z(k), β(k)) and the actual value y(k): e(k)=y(k) h( (k|k 1),z(k),β(k)) S15, calculate the Kalman gain based on the initial estimate of the covariance of the current state estimation error and the covariance of the observation noise.

[0042] For example, the Kalman gain K(k) at the current moment can be calculated using the following formula: K(k) = P(k|k 1)H(k) T [H(k)P(k|k 1)H(k) T + R(k)] 1 Where H(k) is the Jacobian matrix of the observation equation with respect to the state vector, and R(k) is the observation noise covariance. For weakly nonlinear or computationally limited downhole edge computing units, F(k) and H(k) can be implemented using local linearization or lookup table methods.

[0043] S16. The initial estimate of the state vector at the current time is updated using the innovation residual and Kalman gain to obtain the estimated value of the state vector at the current time. The initial estimate of the covariance of the state estimation error at the current time is updated using the Kalman gain to obtain the estimated value of the covariance of the state estimation error at the current time.

[0044] For example, the initial estimate of the state vector at the current moment can be obtained using the following formula. (k|k 1) Update the state vector to obtain an estimate of the current state vector. (k|k): (k|k) = (k|k 1) + K(k)[y(k) h( (k|k 1), z(k), β(k))] In some embodiments of this application, after the state vector prediction at the current moment is completed, the process noise covariance Q(k) and the observation noise covariance R(k) can be adaptively corrected for the state vector prediction at the next moment.

[0045] The initial estimate of the covariance P(k|k) of the state estimation error at the current time can be obtained using the following formula. 1) Update the data to obtain the covariance estimate P(k|k) of the current state estimation error, which will be used for calculating the warning confidence in subsequent steps. P(k|k) = [I K(k)H(k)]P(k|k 1) In one implementation, the residual variance of the innovation residual within the sliding window and the estimation error of the state vector at the current time step can be determined (the difference between the initial estimate and the estimated value of the state vector, i.e., ...). (k|k)- (k|k-1)). The observation noise covariance is adaptively corrected based on the residual variance. When the residual variance continuously exceeds the sensor calibration noise upper limit, the R value of the corresponding observation vector is increased. The process noise covariance is adaptively corrected based on the estimation error value of the current state vector. When the state estimation error continuously shifts within the stable window, the Q value of the corresponding state variable is increased.

[0046] As an example, suppose the variance of the innovation residual e(k) of a certain sensor (such as a right column pressure gauge) consistently exceeds the factory-calibrated noise level within the sliding window, indicating that the sensor may have experienced a temporary malfunction or be subject to strong interference. Therefore, the diagonal element of the observation noise covariance R(k) matrix corresponding to the pressure observation vector can be temporarily increased to reduce its reliability. Similarly, suppose the estimation error of the off-center load moment x2 in the state vector exhibits a long-term unidirectional deviation within the steady-state operating window, indicating that the state model cannot keep up with actual changes. Therefore, the diagonal element of the process noise covariance Q(k) matrix corresponding to the state vector can be increased, allowing the filter to more flexibly follow the measured values ​​and avoid estimation lag.

[0047] Figure 7 This document provides a flowchart of the state reconstruction process for an Extended Kalman Filter (EKF) with online parameter identification, as illustrated in an embodiment of this application. Figure 7 As shown, the extended Kalman filter, in addition to state prediction, covariance prediction, Kalman gain calculation, and state update, may also include online parameter identification. Within each coal mining cycle or preset time window, the recursive least squares method is used to identify β(k)=[a 11 ,a 12 ,...,c 11 ,c 12 ,...,k c ,κ d Updates are performed. In one implementation, to adapt to different supports and working conditions, a recursive least squares method with a forgetting factor can be used, allowing the same method to be transferred to different types of hydraulic supports, different mining heights, and different mine pressure conditions. The forgetting factor ρ can be taken from 0.95 to 0.995. Under stable working conditions, ρ is taken as a larger value to maintain parameter smoothness; after non-stationary abrupt changes or sensor replacement, ρ can be appropriately reduced to improve the adaptation speed. Therefore, the embodiments of this application do not simply use EKF for attitude angle smoothing, but use EKF for internal support state reconstruction, and use online identification to make the model adapt to the differences in support type, coal and rock conditions, and equipment wear.

[0048] Step 105: Determine the imbalance index of the hydraulic support based on the estimated value of the state vector.

[0049] In some embodiments of this application, the imbalance index of the hydraulic support may include: eccentric load index, attitude deviation index, constraint anomaly index, and stiffness degradation auxiliary index. The imbalance index of the hydraulic support can be calculated using the following formula: ψ1=| 2| / M0(k) ψ2=z2 / θ0(k) ψ3=| 4| / F0(k) ψ4=max{0,1 5} / D0 Among them, ψ1 is the eccentric load index, used to evaluate the degree to which the eccentric load moment approaches the design allowable eccentric load moment; ψ2 is the attitude deviation index, used to evaluate the degree of attitude deviation of the top beam relative to the base; ψ3 is the constraint anomaly index, used to evaluate the degree of constraint anomaly between the frames; and ψ4 is the stiffness degradation auxiliary index, used to evaluate the degree of support stiffness degradation. 2 represents the estimated value of the eccentric load moment, M0(k) is the reference value of the eccentric load moment, which can be determined by the rated working resistance, the equivalent force arm, and the statistical quantile of the eccentric load in the historical stability window. z2 represents the attitude deviation component of the top beam relative to the base, and θ0(k) is the attitude deviation reference value, which can be determined by the allowable attitude deviation of the support structure and the attitude fluctuation in the historical stability window. 4 represents the estimated value of the inter-frame constraint reaction force, and F0(k) represents the benchmark of the inter-frame constraint reaction force, which can be determined jointly by the equivalent constraint stiffness between frames, the design step distance, and the inter-frame displacement fluctuation in the historical stability window. 5 is the estimated value of the support stiffness degradation coefficient, and D0 is the stiffness degradation alarm benchmark.

[0050] In one implementation, M0(k), θ0(k), and F0(k) can be determined by a combination of structural design thresholds and historical stability window statistical thresholds. For example, M0(k) = min{M design ,q 0.95 (| 2| stable )·γ M}, θ0(k)=min{θ design ,q 0.95 (|z2| stable )·γ θ}, F0(k)=min{F design ,q 0.95 (| 4| stable )·γ F}, where q 0.95represents the 95th percentile in the stable window, γ M , γ θ , γ F are safety amplification factors.

[0051] Step 106: determining the support state grade of the hydraulic support and issuing an early warning according to the imbalance index and a preset support state grading judgment rule.

[0052] In some embodiments of the present application, the support state grading judgment rules include: when ψ1<T1, ψ2<T1, and ψ3<T1, it is determined as a normal state L0, which indicates that the support state is stable, a heartbeat signal can be output, and no adjustment is required; when ψ1≥T1, ψ2<T1, and ψ3<T1, it is determined as an unbalanced load early warning state L1, which indicates unbalanced development of loads on the left and right upright columns, and it is recommended to adjust the setting load of the front / right upright column; when ψ2≥T1, ψ1<T1, and ψ3<T1, it is determined as a posture early warning state L2, which indicates the development of posture deviation of the top beam relative to the base seat, and it is recommended to correct the posture of the balance jack; when at least one of the following conditions is met, it is determined as a compound imbalance state L3: ψ1≥T1 and ψ2≥T1, ψ1>T2 or ψ2≥T1, and ψ4 is greater than T3; the compound imbalance state L3 indicates compound imbalance of load and posture, and operations including suspending pushing, cooperating with adjacent supports and manual rechecking can be performed; when ψ3≥T1, it is determined as an abnormal constraint state L4, which indicates inter-support extrusion or abnormal pushing constraint, and operations including locking pushing, checking inter-support connection and bottom plate conditions can be performed, and the priority of L4 is higher than that of L0-L3; wherein T1<T2. Optionally, the first threshold T1 can be 0.5, the second threshold T2 can be 0.8, and the stiffness degradation threshold T3 can be 0.6.

[0053] Figure 8 is a schematic diagram of a support imbalance grading early warning decision tree provided by an embodiment of the present application. As Figure 8 shown, in some embodiments of the present application, grading early warning can also be performed by combining the imbalance index and early warning confidence Conf. The early warning confidence Conf can be calculated jointly by the covariance estimation value P(k) of the current time state estimation error, the effective data ratio of the sensor and the innovation residual consistency. Exemplarily, Conf=ω1·exp tr(P norm )]+ω2·R valid +ω3·R res . Wherein, P norm is normalized estimated covariance, R valid is an effective sensor data ratio, R resThe system assigns a residual consistency score. When the warning confidence level (Conf) exceeds the preset confidence threshold, the system outputs an automatic control suggestion; when the warning confidence level (Conf) is low, the system outputs a manual review suggestion.

[0054] Figure 9 This is a diagram illustrating a hierarchical early warning and security control interface output architecture proposed in an embodiment of this application. Figure 9 As shown, in some embodiments of this application, suggested parameters corresponding to the support imbalance level can be output through the safety control interface module. The safety control interface module 680 can communicate with the working face centralized control system 700. The safety control interface module outputs suggested parameters with safety boundaries, rather than unconstrained control commands. For L1 off-center load warning, if the pressure on the right column is lower than that on the left column and the right column has not yet reached its rated working resistance, it is recommended to increase the initial support force of the right column by 3% to 12%; if the pressure on the left column is lower than that on the right column, it is recommended to increase the initial support force of the left column or reduce the attitude deviation of the balancing jack. For L2 attitude warning, it is recommended to activate the balancing jack, and the attitude correction amount shall not exceed the maximum allowable correction angle. For L3 composite imbalance, it is recommended to suspend the pushing of this frame, lock the current support lower limit, notify the adjacent supports for coordinated support, and prompt manual verification. For L4 abnormal constraint, it is recommended to lock the pushing operation, check the inter-frame connection, the bottom plate undulation, and the position of adjacent frames.

[0055] All the above adjustment suggestions meet the following boundary conditions: the adjusted column pressure does not exceed the rated working resistance or the safety valve opening pressure; the adjusted support resistance is not lower than the minimum safe support resistance; the balancing jack movement does not exceed the mechanical allowable stroke and the maximum attitude correction; the coordinated movement of adjacent supports does not cause adjacent supports to enter a higher risk state; when the warning confidence level is lower than the threshold, only manual review suggestions are output to reduce the risk of over-control.

[0056] To avoid the potential warning lag caused by relying solely on steady-state thresholds under non-steady working conditions such as periodic pressure, fault crossing, roof collapse, or localized floor heave, some embodiments of this application can detect sudden changes in non-steady working conditions while simultaneously detecting the support imbalance level. Figure 10 This is a flowchart illustrating a non-stationary operating condition abrupt change detection method proposed in an embodiment of this application. Figure 11 This is a flowchart illustrating another method for graded early warning of imbalance in hydraulic support for ultra-high and ultra-long working faces, provided as an embodiment of this application. Figure 10 and Figure 11 As shown, the estimated value of the state vector can be obtained. 1- 5. Establish a system of length T for the coupled eigenvectors z1-z5 respectively. wA sliding window is used to calculate the mutation detection rate. When the mutation detection rate of the estimated state vector or the mutation detection rate of the coupled feature vector exceeds the corresponding mutation threshold, the hydraulic support is determined to be in a non-stationary mutation state at the current moment, and a corresponding early warning is issued to control and improve the process noise covariance.

[0057] For example, the mutation detection parameters may include mean offset Δμ and variance change rate Δσ. 2 And the cumulative sum statistic (CUSUM). The mutation detection rate of the estimated state vector or the mutation detection rate of the coupled feature vector exceeds the corresponding mutation threshold, which may include: |Δμ|>η1, Δσ 2 >η2 or CUSUM>η3.

[0058] To avoid false alarms due to short-term electrical noise, the above-mentioned sudden change detection conditions can be required to be continuously met for at least two adjacent sampling periods, or simultaneously accompanied by the proportion of effective sensor data exceeding a preset proportion.

[0059] After a non-stationary mutation event is triggered, the process noise covariance Q(k) of the EKF can be increased, allowing the state estimation to track the actual changes more quickly. Simultaneously, the warning flag is output to the working face centralized control system via the safety control interface module. For high-confidence mutation events, the output time can be set to no more than 200ms.

[0060] By implementing the embodiments of this application, multi-source information characterizing the support status is constructed into a coupled feature vector, and a state-space model is established with the internal mechanical state of the support (which cannot be directly measured) as the state vector and measurable physical quantities as the observables. An extended Kalman filter is used to recursively estimate the state vector, thus achieving online reconstruction of the internal support status of the stent. Based on the reconstructed state, an imbalance index is calculated, and early warning is issued according to a tiered judgment rule. This elevates the basis for imbalance identification from threshold judgment of apparent measurable variables to a comprehensive assessment of the internal mechanical state. It can detect early signs of support imbalance before a single sensor reading reaches the traditional alarm threshold, increasing the lead time for anomaly identification and reducing the risk of false alarms and missed alarms caused by sensor noise or instantaneous disturbances.

[0061] To better understand the embodiments of this application, Figure 12 This is a comparison diagram of the early warning effect under a typical working condition proposed in an embodiment of this application. Taking the application of the method of this application in a certain ultra-long fully mechanized mining face as an example, several hydraulic supports in the central area are selected as monitoring objects. Taking a certain support as an example, p is obtained by sampling. f p r θ p θ r θ bp θ br p b Δs(n), Δs(n+1), and d nAfter synchronization, the system calculates z1, z2, z3, z4, and z5, and then reconstructs them using EKF. 1 to 5.

[0062] Before a certain period of pressure, the univariate pressure p f and p r Neither exceeded the traditional pressure alarm threshold, nor did the top beam attitude exceed the traditional attitude alarm threshold; however, z1 continued to rise, and z3 showed a simultaneous increase in both load change rate and attitude change rate, triggering EKF reconstruction. 2. The load exceeds 0.5M0. Based on this, the system determines it as an L1 off-center load warning and outputs a suggestion to adjust the initial support force of the right column. After adjustment, 2. The price fell back, but did not develop into a complex imbalance during the subsequent period of pressure.

[0063] Compared to single-variable threshold monitoring, the early warning triggering basis of this application embodiment is not the exceeding of a single sensor reading, but rather the fact that the coupling characteristics and internal state have reached the pre-imbalance condition. Therefore, when a single pressure or tilt angle has not yet reached the traditional alarm threshold, this invention can output an off-center load warning or attitude warning in advance.

[0064] In the comparative experiment, three baseline schemes can be set: the first group is an alarm based solely on the column pressure threshold; the second group is an alarm based solely on the simple logic thresholds of pressure, tilt angle, and displacement; and the third group is an alarm based solely on ordinary attitude filtering. The scheme of this invention is used as the fourth group. Recommended evaluation indicators include early warning lead time, false alarm rate of imbalance events, false alarm rate, number of manual checks, number of shutdowns due to support abnormalities, and successful effective adjustment rate.

[0065] like Figure 12 As shown, the first single-pressure threshold detection scheme is simple to implement but provides an early warning; The second set of multivariable logic threshold detection schemes cannot reflect attitude-load coupling; The third set of ordinary attitude fusion detection schemes cannot reconstruct the off-center load moment and constraint reaction force; The fourth group of detection schemes proposed in this application can identify imbalance intermediate states in advance and output safety control recommendations.

[0066] Figure 13 This is a schematic diagram of a graded early warning device for imbalance of hydraulic support for ultra-high and ultra-long working faces, provided as an embodiment of this application. Figure 13 As shown, the hydraulic support imbalance graded early warning device for ultra-high and ultra-long working faces may include: acquisition module 131, feature extraction module 132, construction module 133, solution module 134, determination module 135 and graded early warning module 136.

[0067] The acquisition module 131 is used to acquire multi-source information of the hydraulic support for the ultra-high and ultra-long working face. The multi-source information includes: left column pressure, right column pressure, top beam pitch angle, top beam tilt angle, base pitch angle, base tilt angle, balance jack pressure, pushing displacement, adjacent support pushing displacement and relative displacement between supports. Feature extraction module 132 is used to construct coupled feature vectors based on multi-source information. The coupled feature vectors include: column load asymmetric component, top beam relative to base attitude deviation component, attitude-load coordinated change rate component, inter-frame constraint component, and load disturbance component. Module 133 is constructed to establish a support state space model that includes state equations and observation equations, using the internal support resultant force, eccentric load moment, attitude-load coupling degree, inter-frame constraint reaction force and support stiffness degradation coefficient as state vectors, and column pressure, eccentric load equivalent moment, relative attitude deviation, push-constraint coupling displacement and balancing jack pressure as observation vectors. The solver module 134 is used to recursively solve the support state space model based on multi-source information and coupled eigenvectors using an extended Kalman filter, and obtain the estimated value of the state vector and the estimated value of the covariance of the state estimation error. Module 135 is used to determine the imbalance index of the hydraulic support based on the estimated value of the state vector; The graded early warning module 136 is used to determine the support status level of the hydraulic support and issue an early warning based on the imbalance index and the preset support status grading judgment rules.

[0068] In some embodiments of this application, the solving module 134 is specifically used for: calculating the initial estimate of the state vector at the current time based on the state equation and the state vector at the previous time; calculating the initial estimate of the covariance of the state estimation error at the current time based on the covariance of the state estimation error at the previous time and the process noise covariance; determining the theoretical and actual values ​​of the observation vector based on the multi-source information, the coupled feature vector, and the observation equation; determining the innovation residual based on the theoretical and actual values ​​of the observation vector; calculating the Kalman gain based on the initial estimate of the covariance of the state estimation error at the current time and the observation noise covariance; updating the initial estimate of the state vector at the current time using the innovation residual and the Kalman gain to obtain the estimate of the state vector at the current time, and updating the initial estimate of the covariance of the state estimation error at the current time using the Kalman gain to obtain the estimate of the covariance of the state estimation error at the current time.

[0069] In some embodiments of this application, the solver module 134 is further configured to: calculate the residual variance of the sliding window and the estimation error value of the state vector at the current time based on the innovative residual; adaptively correct the observation noise covariance based on the residual variance; and adaptively correct the process noise covariance based on the estimation error value of the state vector at the current time.

[0070] In some embodiments of this application, such as Figure 13 Based on the illustrated embodiment, the hydraulic support imbalance classification and early warning device for ultra-high and ultra-long working faces may further include a sudden change detection module. The sudden change detection module is used to: establish sliding windows to calculate the sudden change detection quantity for the estimated value of the state vector and the coupled feature vector respectively; when the sudden change detection quantity of the estimated value of the state vector or the sudden change detection quantity of the coupled feature vector exceeds the corresponding sudden change threshold, it determines that the hydraulic support is in a non-stationary sudden change state at the current moment and issues a corresponding early warning, thereby controlling and improving the process noise covariance.

[0071] To implement the above embodiments, this application also proposes an electronic device, including: a processor and a memory communicatively connected to the processor; the memory stores computer execution instructions; the processor executes the computer execution instructions stored in the memory to implement the method provided in the foregoing embodiments.

[0072] To implement the above embodiments, this application also proposes a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the methods provided in the foregoing embodiments.

[0073] To implement the above embodiments, this application also proposes a computer program product, including a computer program that, when executed by a processor, implements the methods provided in the foregoing embodiments.

[0074] In the foregoing descriptions of the embodiments, the terms "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0075] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0076] Any process or method description in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more executable instructions for implementing custom logic functions or processes, and the scope of the preferred embodiments of this application includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as should be understood by those skilled in the art to which embodiments of this application pertain.

[0077] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of computer-readable media include: an electrical connection having one or more wires (electronic device), a portable computer disk drive (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Alternatively, the computer-readable medium may be paper or other suitable media on which the program can be printed, since the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in a computer memory.

[0078] It should be understood that various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0079] Those skilled in the art will understand that all or part of the steps of the methods in the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, the program includes one or a combination of the steps of the method embodiments.

[0080] Furthermore, the functional units in the various embodiments of this application can be integrated into a processing module, or each unit can exist physically separately, or two or more units can be integrated into a module. The integrated module can be implemented in hardware or as a software functional module. If the integrated module is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium.

[0081] The storage medium mentioned above can be a read-only memory, a disk, or an optical disk, etc. Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of this application.

Claims

1. A method for graded early warning of imbalance in hydraulic support for ultra-high and ultra-long working faces, characterized in that, Includes the following steps: Acquire multi-source information for hydraulic supports used in ultra-high and ultra-long working faces. The multi-source information includes: left column pressure, right column pressure, top beam pitch angle, top beam tilt angle, base pitch angle, base tilt angle, balance jack pressure, pushing displacement, adjacent support pushing displacement, and relative displacement between supports. Based on the multi-source information, a coupled feature vector is constructed, which includes: column load asymmetric component, top beam relative to base attitude deviation component, attitude-load coordinated change rate component, inter-frame constraint component, and load disturbance component. Using the internal support resultant force, eccentric load moment, attitude-load coupling degree, inter-frame constraint reaction force and support stiffness degradation coefficient as state vectors, and the column pressure, eccentric load equivalent moment, relative attitude deviation, shoving-constraint coupling displacement and balancing jack pressure as observation vectors, a support state space model including state equations and observation equations is established. An extended Kalman filter is used to recursively solve the support state space model based on the multi-source information and the coupling feature vector to obtain the estimated value of the state vector and the covariance estimate of the state estimation error. The imbalance index of the hydraulic support is determined based on the estimated value of the state vector; Based on the imbalance index and the preset support status grading rules, the support status level of the hydraulic support is determined and an early warning is issued.

2. The method according to claim 1, characterized in that, The coupled feature vector is constructed using the following formula: z1=|p f p r | / p rated z2=√[(θ p i bp ) 2 +(θ r i br ) 2 ] / θ ref z3=α1·|dz1 / dt|+α2·|dz2 / dt|+α3·|d(p b / p rated ) / dt| z4=[|Δs(n) Δs(n+1)|+κ d ·|d n |] / Δs design z5=σ τ (p f +p r ) / (2p rated ) Wherein, z1 is the asymmetric component of the column load, z2 is the attitude deviation component of the top beam relative to the base, z3 is the attitude-load coordinated rate of change component, z4 is the inter-frame constraint component, z5 is the load disturbance component, and p f p is the pressure on the left column. r θ represents the pressure on the right column. p Let θ be the pitch angle of the top beam. r Let θ be the tilt angle of the top beam. bp Let θ be the pitch angle of the base. br p is the tilt angle of the base. b The balancing jack pressure is denoted by Δs(n), the displacement is denoted by Δs(n+1), and the displacement of the adjacent support is denoted by d. n p represents the relative displacement between the frames. rated For the rated column pressure, θ ref As the baseline for attitude tolerance, Δs design The design step size is τ, where τ is the length of the sliding window, and σ is the sliding window size. τ (·) represents the standard deviation within the window τ, where α1, α2, α3, and κ are the standard deviations. d These are calibrable weighting coefficients.

3. The method according to claim 1, characterized in that, The support state-space model is recursively solved based on the multi-source information and the coupled feature vector, including: Based on the state equation and the state vector at the previous time step, calculate the initial estimate of the state vector at the current time step; Based on the covariance of the state estimation error at the previous time step and the process noise covariance, calculate the initial estimate of the covariance of the state estimation error at the current time step. Based on the multi-source information, the coupled feature vector, and the observation equation, the theoretical and actual values ​​of the observation vector are determined. The innovation residual is determined based on the theoretical and actual values ​​of the observed vector; The Kalman gain is calculated based on the initial estimate of the covariance of the current state estimation error and the observation noise covariance. The initial estimate of the state vector at the current time is updated using the innovation residual and the Kalman gain to obtain the estimated value of the state vector at the current time. The initial estimate of the covariance of the state estimation error at the current time is updated using the Kalman gain to obtain the estimated value of the covariance of the state estimation error at the current time.

4. The method according to claim 3, characterized in that, Also includes: The residual variance of the sliding window and the estimation error of the state vector at the current moment are calculated based on the innovative residual calculation. The observation noise covariance is adaptively corrected based on the residual variance. The process noise covariance is adaptively corrected based on the estimation error value of the state vector at the current moment.

5. The method according to claim 1, characterized in that, The imbalance index of the hydraulic support includes: eccentric load index, attitude deviation index, constraint anomaly index, and stiffness degradation auxiliary index; the imbalance index of the hydraulic support is calculated using the following formula: ψ1=| 2| / M0(k) ψ2=z2 / θ0(k) ψ3=| 4| / F0(k) ψ4=max{0,1 5} / D0 Wherein, ψ1 is the off-center loading index, ψ2 is the attitude deviation index, ψ3 is the constraint anomaly index, and ψ4 is the stiffness degradation auxiliary index. 2 represents the estimated value of the eccentric load moment, M0(k) is the reference value of the eccentric load moment, z2 is the attitude deviation component of the top beam relative to the base, and θ0(k) is the attitude deviation reference value. 4 represents the estimated value of the inter-frame constraint reaction force, and F0(k) represents the reference value of the inter-frame constraint reaction force. 5 is the estimated value of the support stiffness degradation coefficient, and D0 is the stiffness degradation alarm benchmark.

6. The method according to claim 5, characterized in that, The rules for determining the support status include: When ψ1<T1, ψ2<T1 and ψ3<T1, it is determined as a normal state L0; When ψ1≥T1, ψ2<T1 and ψ3<T1, it is determined as an unbalanced load early warning state L1; When ψ2≥T1, ψ1<T1 and ψ3<T1, it is determined as a posture early warning state L2; When at least one of the following conditions is satisfied, it is determined as a compound unbalance state L3: ψ1≥T1 and ψ2≥T1, ψ1>T2 or ψ2≥T1, ψ4 is greater than T3; When ψ3≥T1, it is determined as an abnormal constraint state L4; wherein T1<T2.

7. The method according to claim 3, characterized in that, The method further comprises: establishing sliding windows for the estimated value of the state vector and the coupling feature vector respectively to calculate a mutation detection quantity; when the mutation detection quantity of the estimated value of the state vector or the mutation detection quantity of the coupling feature vector exceeds a corresponding mutation threshold, determining that the hydraulic support at the current moment is in a non-stationary mutation state, giving a corresponding early warning, and controlling to increase the process noise covariance.

8. A graded early warning device for imbalance of hydraulic support in ultra-high and ultra-long working faces, characterized in that, The device comprises: an acquisition module, configured to acquire multi-source information of a hydraulic support for an ultra-high and ultra-long working face, wherein the multi-source information comprises: left column pressure, right column pressure, top beam pitch angle, top beam roll angle, base pitch angle, base roll angle, balance jack pressure, pushing displacement, adjacent support pushing displacement and inter-support relative displacement; a feature extraction module, configured to construct a coupling feature vector based on the multi-source information, wherein the coupling feature vector comprises: column load asymmetric component, top beam relative base posture deviation component, posture-load collaborative change rate component, inter-support constraint component and load disturbance component; a construction module, configured to take internal supporting resultant force, unbalanced load moment, posture-load coupling degree, inter-support constraint reaction force and supporting stiffness degradation coefficient as a state vector, and take column pressure, unbalanced load equivalent moment, relative posture deviation, pushing-constraint coupling displacement and balance jack pressure as an observation vector, to establish a supporting state space model comprising a state equation and an observation equation; a solving module, configured to adopt an extended Kalman filter to perform recursive solution on the supporting state space model based on the multi-source information and the coupling feature vector, to obtain an estimated value of the state vector and a covariance estimated value of a state estimation error; a determining module, configured to determine an unbalance index of the hydraulic support based on the estimated value of the state vector; a graded early warning module, configured to determine a supporting state grade of the hydraulic support and perform early warning according to the unbalance index and a preset supporting state grading determination rule.

9. An electronic device, characterized in that, The system comprises: a processor, and a memory communicatively connected to the processor; the memory stores computer execution instructions; the processor executes the computer execution instructions stored in the memory, so as to implement the method according to any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, the computer-readable storage medium stores computer execution instructions, and when the computer execution instructions are executed by a processor, the processor is configured to implement the method according to any one of claims 1-7.

Images